A self-supervised method segments time series by predictive surprise from a frozen forecaster, clusters segments into dynamical concepts, and classifies with reliability-aware abstention, achieving competitive accuracy and interpretability without annotations.
Time-series classifiers deployed in risk-sensitive domains must be (1) accurate, (2) interpretable, and (3) able to abstain when evidence is weak. Existing methods do not deliver all 3 at once, and the obstacle is structural, rather than logistical. Meaningful temporal patterns cannot be defined without knowing segment boundaries, yet meaningful boundaries cannot be placed without knowing which patterns to expect. Existing approaches break this circularity by using fixed windows or expert-annotated vocabularies, thereby sacrificing at least one of the 3 properties. We introduce ConceptTime, which closes the loop through a single self-supervised signal. A frozen probabilistic forecaster…
time-series classifiersself-supervised signalfrozen probabilistic forecasterpredictive surprisesegment boundariesdynamical regimesconcept sequencelightweight headreliability by designOOD detection
- Upvotes
- 3
- Authors
- 5
- Published
- Nov 7, 2026
Summary and abstract excerpt via Hugging Face Papers ↗ · arXiv.